Re: [rtcweb] Requesting "SDP or not SDP" debate to be re-opened

[Harald Alvestrand <[e-mail]>]

On 06/20/2013 06:15 PM, Iñaki Baz Castillo wrote:
> 2013/6/20 Harald Alvestrand <[e-mail]>:
>> I think interoperability with an application written to the implementations
>> that are based on the current specifications (that is - a Javascript
>> application that you could load into both Chrome and your demo of CU-RTCWeb,
>> using a JS shim in your demo, and have them talk to each other) would be a
>> pretty compelling argument.
> Hi Harald, with all due respect:
>
> So, it seems that the current SDP-based spec is God, and if somebody
> wants to propose another alternative, it must be compatible with God,
> regardless that God is not yet a proven standard and what some of us
> are asserting is that God is bad, very bad. Am I right?

Inaki, that's quite a bit below your usual standard of argument.

>
> IMHO it would be much more interesting a JS demo that takes the SDP
> generated by Chrome and "converts" it into a Jingle XML session
> description (I mean XEP-0167), and that can interoperate (via XMPP
> over WebSocket or whatever) with a XMPP/Jingle endpoint by
> incrementally sending/receiving transport (ICE stuff) and media
> capabilities (as XEP-0167 allows) in different XMPP messages.

You find it interesting. I'm sure you have a particular XMPP/Jingle
endpoint in mind that you want to connect to.

Create one. Or try, and share with us why it did not work.


>
> And it would also be interesting a JS demo that interoperates with a
> SIP gateway and is able to perform the "hold" / "unhold" feature so
> widely extended in SIP world (via reINVITE with a=inactive/sendonly
> and 200 "OK" with a=inactive/recvonly).

We have people who have demonstrated interoperability with SIP softphones.
Some of them are on this list.
Can they tell us what happened when they tried?

>
> It seems that we must prove how bad SDP is while the fact is that
> those in favour of the SDP-based model should demonstrate how
> "flexible" and powerful it is. As far as we can see in the spec
> development, SDP does not seem to be so good and hence it should be
> questionable IMHO.

The purported value of SDP is interoperability (and the idea that specs
defining what it's supposed to mean already exist).
If you can show (by code) that this purported value can be achieved
without SIP, you have an argument.

If you cannot show anything, you have words.


>
> Just my opinion.
>
> Best regards.
>
> --
> Iñaki Baz Castillo
> <[e-mail]>